Search

Metabolomics has gained much attention due to its potential to reveal molecular disease mechanisms and present viable biomarkers. This work uses a panel of untargeted serum metabolomes from 602 children from the COPSAC2010 mother–child cohort. The annotated part of the metabolome consists of 517 chemical compounds curated using automated procedures. We created a filtering method for the quantified metabolites using predicted quantitative structure–bioactivity relationships for the Tox21 database on nuclear receptors and stress response in cell lines. The metabolites measured in the children’s serums are predicted to affect specific targeted models, known for their significance in inflammation, immune function, and health outcomes. The targets from Tox21 have been used as targets with quantitative structure–activity relationships (QSARs). They were trained for ~7000 structures, saved as models, and then applied to the annotated metabolites to predict their potential bioactivities. The models were selected based on strict accuracy criteria surpassing random effects. After application, 52 metabolites showed potential bioactivity based on structural similarity with known active compounds from the Tox21 set. The filtered compounds were subsequently used and weighted by their bioactive potential to show an association with early childhood hs-CRP levels at six months in a linear model supporting a physiological adverse effect on systemic low-grade inflammation.

Vertical transmission of metabolic constituents from mother to child contributes to the manifestation of disease phenotypes in early life. This study probes the vertical transmission of metabolites from mothers to offspring by utilizing machine learning techniques to differentiate between true mother–child dyads and randomly paired non-dyads. Employing random forests (RF), light gradient boosting machine (LGBM), and logistic regression (Elasticnet) models, we analyzed metabolite concentration discrepancies in mother–child pairs, with maternal plasma sampled at 24 weeks of gestation and children’s plasma at 6 months. The propensity of vertical transfer was quantified, reflecting the likelihood of accurate mother–child matching. Our findings were substantiated against an external test set and further verified through statistical tests, while the models were explained using permutation importance and SHapley Additive exPlanations (SHAP). The best model was achieved using RF, while xenobiotics were shown to be highly relevant in transfer. The study reaffirms the transmission of certain metabolites, such as perfluorooctanoic acid (PFOA), but also reveals additional insights into the maternal influence on the child’s metabolome. We also discuss the multifaceted nature of vertical transfer. These machine learning-driven insights complement conventional epidemiological findings and offer a novel perspective on using machine learning as a methodology for understanding metabolic interactions.

Non-invasive detection of unstable angina (UA) patients with different severity of coronary lesions remains challenging. This study aimed to identify plasma lipoproteins (LPs) that can be used as potential biomarkers for assessing the severity of coronary lesions, determined by the Gensini score (GS), in UA patients. We collected blood plasma from 67 inpatients with angiographically normal coronary arteries (NCA) and 230 UA patients, 155 of them with lowGS (GS ≤ 25.4) and 75 with highGS (GS > 25.4), and analyzed it using proton nuclear magnetic resonance spectroscopy to quantify 112 lipoprotein variables. In a logistic regression model adjusted for four well-known risk factors (age, sex, body mass index and use of lipid-lowering drugs), we tested the association between each lipoprotein and the risk of UA. Combined with the result of LASSO and PLS-DA models, ten of them were identified as important LPs. The discrimination with the addition of selected LPs was evaluated. Compared with the basic logistic model that includes four risk factors, the addition of these ten LPs concentrations did not significantly improve UA versus NCA discrimination. However, thirty-two selected LPs showed notable discrimination power in logistic regression modeling distinguishing highGS UA patients from NCA with a 14.9% increase of the area under the receiver operating characteristics curve. Among these LPs, plasma from highGS patients was enriched with LDL and VLDL subfractions, but lacked HDL subfractions. In summary, we conclude that blood plasma lipoproteins can be used as biomarkers to distinguish UA patients with severe coronary lesions from NCA patients.

The pregnancy period and first days of a newborn’s life is an important time window to ensure a healthy development of the baby. This is also the time when the mother and her baby are exposed to the same environmental conditions and intake of nutrients, which can be determined by assessing the blood metabolome. For this purpose, dried blood spots (DBS) of newborns are a valuable sampling technique to characterize what happens during this important mother-child time window. We used metabolomics profiles from DBS of newborns (age 2–3 days) and maternal plasma samples at gestation week 24 and postpartum week 1 from n=664 mother-child pairs of the Copenhagen Prospective Studies on Asthma in Childhood 2010 (COPSAC2010) cohort, to study the vertical mother-child transfer of metabolites. Further, we investigated how persistent the metabolites are from the newborn and up to 6 months, 18 months, and 6 years of age. Two hundred seventy two metabolites from UPLC-MS (Ultra Performance Liquid Chromatography-Mass Spectrometry) analysis of DBS and maternal plasma were analyzed using correlation analysis. A total of 11 metabolites exhibited evidence of transfer (R>0.3), including tryptophan betaine, ergothioneine, cotinine, theobromine, paraxanthine, and N6-methyllysine. Of these, 7 were also found to show persistence in their levels in the child from birth to age 6 years. In conclusion, this study documents vertical transfer of environmental and food-derived metabolites from mother to child and tracking of those metabolites through childhood, which may be of importance for the child’s later health and disease.

The relationship between developmental milestone achievement in infancy and later cognitive function and mental health is well established, but underlying biochemical mechanisms are poorly described. Our study aims to discover pathways connected to motor milestone achievement during infancy by using untargeted plasma metabolomic profiles from 571 six-month-old children in connection with age of motor milestones achievement (Denver Developmental Index) in the Copenhagen Prospective Studies on Asthma in Childhood 2010 (COPSAC2010) mother–child cohort. We used univariate regression models and multivariate modelling (Partial Least Squares Discriminant Analysis: PLS-DA) to examine the associations and the VDAART (Vitamin D Antenatal Asthma Reduction Trial) cohort for validation. The univariate analyses showed 62 metabolites associated with gross-motor milestone achievement (p < 0.05) as well as the PLS-DA significantly differentiated between slow and fast milestone achievers (AUC = 0.87, p = 0.01). Higher levels of tyramine-O-sulfate in the tyrosine pathway were found in the late achievers in COPSAC (p = 0.0002) and in VDAART (p = 0.02). Furthermore, we observed that slow achievers were characterized by higher levels of fatty acids and products of fatty acids metabolism including acyl carnitines. Finally, we also observed changes in the lysine, histidine, glutamate, creatine and tryptophan pathways. Observing these metabolic changes in relation to gross-motor milestones in the first year of life, may be of importance for later cognitive function and mental health.